spatial and temporal factors controlling short … · 3 research group ecosystem management,...
TRANSCRIPT
SEDIMENTATION IN TIDAL MARSHES 739
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Earth Surface Processes and Landforms
Earth Surf. Process. Landforms 28, 739–755 (2003)Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/esp.495
SPATIAL AND TEMPORAL FACTORS CONTROLLING SHORT-TERM
SEDIMENTATION IN A SALT AND FRESHWATER TIDAL MARSH,
SCHELDT ESTUARY, BELGIUM, SW NETHERLANDS
S. TEMMERMAN1*, G. GOVERS1, S. WARTEL2 AND P. MEIRE3
1 Laboratory for Experimental Geomorphology, KU Leuven, Redingenstraat 16, B-3000 Leuven, Belgium2 Sedimentology Department, Royal Belgian Institute of Natural Sciences, Vautierstraat 29, B-1000 Brussels, Belgium
3 Research Group Ecosystem Management, University of Antwerp, Universiteitsplein 1-c, B-2610 Antwerp, Belgium
Received 20 May 2002; Revised 3 January 2003; Accepted 17 January 2003
ABSTRACT
During a one-year period temporal and spatial variations in suspended sediment concentration (SSC) and deposition werestudied on a salt and freshwater tidal marsh in the Scheldt estuary (Belgium, SW Netherlands) using automatic watersampling stations and sediment traps. Temporal variations were found to be controlled by tidal inundation. The initial SSC,measured above the marsh surface at the beginning of inundation events, increases linearly with inundation height at hightide. In accordance with this an exponential relationship is observed between inundation time and sedimentation rates,measured over 25 spring–neap cycles. In addition both SSC and sedimentation rates are higher during winter than duringsummer for the same inundation height or time. Although spatial differences in vegetation characteristics are large betweenand within the studied salt and freshwater marsh, they do not affect the spatial sedimentation pattern. Sedimentation rateshowever strongly decrease with increasing (1) surface elevation, (2) distance from the nearest creek or marsh edge and(3) distance from the marsh edge measured along the nearest creek. Based on these three morphometric parameters, thespatio-temporal sedimentation pattern can be modelled very well using a single multiple regression model for both the saltand freshwater marsh. A method is presented to compute two-dimensional sedimentation patterns, based on spatial implemen-tation of this regression model. Copyright © 2003 John Wiley & Sons, Ltd.
KEY WORDS: saltmarsh; freshwater marsh; suspended sediment concentration; sediment deposition; Schelde river
INTRODUCTION
Within the estuarine and coastal environment, tidal marshes play an important role as essential habitats for plants
and animals and as sinks and/or sources for nutrients, pollutants and sediments (Allen, 2000). These functions
of tidal marshes are strongly affected by sedimentation and changes in marsh surface elevation, whether this is
in balance with relative sea level rise or not. Much attention has been paid to the quantification of sedimentation
rates on tidal marshes, and especially to the question as to whether or not marsh sedimentation will be able to
keep up with sea level rise. A wide range of measuring techniques have been used to quantify marsh sedimen-
tation rates, over time-scales from one single tidal cycle up to several hundreds of years, including sediment
traps (e.g. Reed, 1989; French et al., 1995; Leonard, 1997; Allen and Duffy, 1998b), artificial or natural marker
horizons (e.g. French and Spencer, 1993; Roman et al., 1997), sedimentation–erosion tables (e.g. Cahoon et al.,
2000), and dating of sediment cores using palaeoenvironmental, radiometric or geochemical techniques (e.g.
Cundy and Croudace, 1996; Roman et al., 1997).
Only a few studies have addressed both spatial and temporal variations in contemporary marsh sedimentation
and the physical processes controlling these variations, although such studies are extremely important to under-
stand the basic mechanisms of tidal marsh sedimentation. Furthermore, sedimentation processes were studied
mainly on salt marshes. Studies on freshwater tidal marshes are very sparse and have been carried out mainly in
US marshes (e.g. Orson et al., 1990; Pasternack and Brush, 2001; Neubauer et al., 2002), while data from NW
European freshwater tidal marshes are lacking. Some studies on salt marshes reported that temporal variations
* Correspondence to: S. Temmerman, Laboratory for Experimental Geomorphology, KU Leuven, Redingenstraat 16, B-3000 Leuven,Belgium. E-mail: [email protected]
740 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
in sedimentation rates are mainly controlled by tidal inundation height (Allen and Duffy, 1998a; Christiansen
et al., 2000). Others indicated that wind–wave activity is the dominant controlling factor, leading in some cases
to reduced sedimentation or even erosion (Pethick, 1992; Van Proosdij et al., 2000), but in other places to
increased sediment inputs and consequently higher sedimentation rates (Reed, 1989; Leonard et al., 1995). In
addition seasonal patterns are reported and these are often attributed to variations in biological activity (Hutchinson
et al., 1995; Leonard et al., 1995; Leonard, 1997; Pasternack and Brush, 2001). Spatial sedimentation patterns
seem to be related to several parameters like marsh surface elevation (e.g. Stoddart et al., 1989; Cahoon and
Reed, 1995), the tidal creek network (e.g. French and Spencer, 1993; French et al., 1995; Leonard, 1997; Reed
et al., 1999), and differences in vegetation structure (Leonard et al., 1995; Leonard, 1997; Boorman et al., 1998).
However, the relative importance and interactions between the different variables thought to control temporal
and spatial variations in marsh sedimentation rates are poorly understood. As a consequence, these overall spatial
and temporal variations are difficult to predict.
This paper presents a detailed study on the spatial and temporal sedimentation patterns in two contrasting
marsh types within the Scheldt estuary, a salt and freshwater tidal marsh. First, field measurements are carried
out to identify the relative importance of the various factors controlling spatial and temporal variations in
sedimentation rates. Secondly, it was investigated to what extent both spatial and temporal variations can be
correctly predicted using a relatively simple, topographically based model that integrates the effects of the
different controlling variables. Finally, special attention is given to whether sedimentation patterns are different
within the studied salt and freshwater marsh.
THE STUDY AREA
The Scheldt estuary (e.g. Meire et al., 1992), situated in the southwest of the Netherlands and the northwest of
Belgium (Figure 1), is characterized by a semidiurnal, meso- to macrotidal regime. The mean tidal range at the
mouth in the southern North Sea ranges between 4·46 and 2·97 m during spring and neap tides respectively
(Claessens and Meyvis, 1994). As the tidal wave enters the estuary, these mean tidal ranges increase to 5·93 m
and 4·49 m at Schelle and then decrease further inland to 2·24 m and 1·84 m near Ghent. The freshwater
discharge of the Scheldt catchment varies between 50 m3 s−1 during dry summer and 300 m3 s−1 during wet
winter months (Taverniers, 2000). Its influence on water levels is only significant at the inland border of the
Figure 1. The Scheldt estuary: (A) location within western Europe; (B) location of salt, brackish and freshwater tidal marshes together withthe study areas, the Paulina and Notelaar marsh
SEDIMENTATION IN TIDAL MARSHES 741
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
estuary and decreases rapidly seaward. Strong northwesterly winds can create large storm surges with resulting
high water levels 2 to 3 m higher than the mean high water level (Claessens and Meyvis, 1994). Wave action
due to wind is expected to decrease landward from the mouth, as a consequence of declining fetch distances
along the estuary.
The suspended sediment concentration (SSC) in the stream channel of the Scheldt estuary typically varies in
time and space, both longitudinally along the estuary and vertically within the water column (Wartel, 1973,
1977). The SSC in the upper part of the water column (which floods the tidal marshes) varies along the estuary
from 30–60 mg l−1 between the mouth and the Dutch-Belgian border up to 100–200 mg l−1 between the border
and Temse (Van Eck et al., 1991; Van Damme et al., 2001). Further upstream the SSC again decreases to 50–
100 mg l−1. Large temporal variations in SSC however occur, depending on semidiurnal, spring–neap and sea-
sonal variations in tidal range and fresh water discharge (e.g. Fettweis et al., 1998).
Along the Scheldt estuary a full salinity gradient exists from salt to fresh water. As a consequence the tidal
marshes in the estuary range from salt marshes, with typical halophytic vegetation, over brackish marshes, with
partially halophytic/helophytic plant species, to freshwater tidal marshes, which are covered only with helophytes
(Figure 1B) (Van den Bergh et al., 2001). Between these marsh types there is a remarkable difference in vegeta-
tion height, which ranges from maximum 0·4–0·8 m on the salt marshes up to 4 m on the freshwater marshes.
Along these estuarine gradients, two contrasting study areas were chosen: (1) the Paulina marsh, a salt
marsh near the mouth where average SSC (around 50 mg l−1 near the water surface) and mean tidal range (3·9 m)
are lowest; (2) the Notelaar marsh, a freshwater tidal marsh near Temse with highest average SSC (100–
200 mg l−1) and tidal range (5·3 m) (Figure 1B). Both study sites are similar in surface area and geomorphology,
typically consisting of a vegetated marsh platform, dissected by networks of tidal creeks that narrow and shallow
inland. The most visible contrasting element is the marsh vegetation. The Paulina marsh is overgrown with
typical NW European salt marsh species, such as Puccinellia maritima, Aster tripolium and Atriplex portulacoides
in high interior marsh basins and mainly Elytrigia pungens on the natural levees. In front of the high Paulina
marsh, a lower marsh exists which is typically dominated by Spartina townsendii (Figure 2b) (Houtekamer,
1996). On the contrary, the Notelaar marsh has a typical freshwater marsh canopy, with a community of Phra-
gmites australis in the lower parts of the marsh and a community of Salix sp. in the higher parts (Figure 2a)
(Hoffmann, 1993).
METHODOLOGY
Field sampling sites
In order to study the impact of spatial factors on the sedimentation pattern, permanent sampling sites were
established along a series of transects covering the main geomorphic units and vegetation types on the salt and
freshwater marsh (Figure 2 and Table I). Three transects were chosen perpendicular to three similar first-order
marsh creeks, containing one measuring site on the natural levee, bordering the creek, and two sites in the lower
inner marsh basin, at a distance of 20 and 40 m from the creek. One transect was established in a typical high
salt marsh canopy (sites 10, 11, 12), and two transects in the two dominating freshwater vegetation types,
Phragmites australis (sites 1, 2, 3) and Salix (sites 7, 8, 9). Another two transects were established over the
whole width of the marsh perpendicular to the marsh edge, both on the salt marsh (sites 13 to 17) and freshwater
marsh (sites 4 to 6). On the salt marsh this transect contains sampling sites on the high marsh as well as on the
lower Spartina marsh. All transects were surveyed relative to Belgian Ordnance Level (TAW, which is approxi-
mately 2·3 m below mean sea level at the Belgian coast), using an electronic total station (Sokkia SET5F). The
sites were further described for their vegetative characteristics (plant species and, where possible, stem density
and height) and grain size characteristics of surface sediments, sampled with metal rings (0·05 m in diameter
and height) and analysed following the standard sieve-pipette method (Table I).
Measuring sediment supply and deposition
Temporal variations in overmarsh suspended sediment concentrations were measured during a one-year period
(from April 2000 until May 2001) from an automatic sampling platform located in a central marsh basin in both
742 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 2. Maps showing vegetative cover and sampling sites at (a) the Notelaar marsh and (b) the Paulina marsh
studied marshes (site 1, 10; Figure 2). For every tidal inundation, the water level above the marsh surface was
recorded every 5 min, using an ISCO flowmeter 4220, and 1 litre water samples were pumped up from 0·15 m
above the marsh surface, using an ISCO sampler 6700. For each inundation cycle a first sample was automat-
ically taken once the inundation height exceeded 0·15 m. Subsequent samples were taken every 30 min, until
the marsh was no longer submerged. Every 15 days (at every neap tide) the filled bottles were collected and new
empty ones were placed in the sampler. To determine the suspended sediment concentration (SSC in g l−1),
the water samples were filtered in the laboratory with preweighed filter papers (pore diameter 0·45 µm), which
were subsequently washed through with deionized water to remove salts. Samples of only four or five inundation
events were analysed for each spring–neap cycle, so that the full range of low to maximum inundation events
during that spring–neap cycle was covered. In all, 245 samples were analysed, covering 27 per cent of all
inundation events during the measuring period.
SEDIMENTATION IN TIDAL MARSHES 743
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Table I. Description of measuring sites for geomorphic situation, dominant plant species, stem density and height (at the endof the growing season), mean sand and clay content of surface sediment
Site Geomorphology* Dominant plant species Stem density† (m−2) Stem height (m) Sand (%) Clay (%)
1,2,4,5 Ba Phragmites australis 90–140 3·7–4·0 1·2 28·93 Le Impatiens glandulifera 60–90 2·2–2·9 1·1 32·96,7 Le Salix sp. – >4·0 25·6 26·78,9 Ba Salix sp. – >4·0 1·4 44·1
10,11,14 Ba Puccinellia maritima, 0·1–0·3 2·5 31·2Atriplex portulacoides – 0·2–0·5
12 Le Elytrigia pungens – 0·3–0·7 51·4 17·013 Ba Elytrigia pungens – 0·3–0·7 1·7 26·215 Lo Aster tripolium, 40–50 0·4–0·7 16·5 23·4
Salicornia 180–240 0·1–0·316,17 Lo Spartina townsendii 400–600 0·4–0·6 11·4 31·5
* Geomorphic units: Ba, inner basin; Le, levee; and Lo, low marsh without basin–levee morphology.
† –, Stem density was not estimated given the nature of vegetation cover (grass, trees).
During the same one-year period, we also sampled the sediment that settled out from suspension on the marsh
surface using on all sampling sites circular plastic sediment traps (diameter 0·233 m). The traps were attached
to the marsh surface using steel claws and were constructed with a floatable cover to protect the deposited
sediment from splash by raindrops during low tides. Every 15 days (at neap tide after each spring–neap cycle)
the traps were collected and replaced by clean ones. In the laboratory, the deposited sediment was washed from
the traps and rinsed with deionized water, to remove salts, and sieved at 707 µm, to remove macroscopic plant
and/or shell material that floated and deposited on the traps. The remaining sediment was then oven-dried at
105 °C and weighed to determine the deposition rate of suspended sediment (in g m−2). In all, 425 samples were
analysed, covering all 25 spring–neap cycles during the year and all 17 sampling sites.
Data assessment and analyses
The water surface was assumed to be horizontal at high tide, so that for each inundation event and every
sampling site maximum inundation height was calculated based on the water level measurements at sites 1 and
10 on the Paulina and Notelaar marsh respectively, and considering the elevation differences between the sites.
Corresponding inundation time was calculated using the observed relationship between maximum inundation
height and time (R2 = 0·89 and 0·96 for the Paulina and Notelaar marsh respectively). By adding up inundation
times of individual inundation events, cumulative inundation times were calculated for each spring–neap cycle.
Since cumulative inundation time reflects both the magnitude and frequency of inundations during a spring–neap
cycle, this parameter was found to be the best to characterize tidal marsh inundation during this time period,
over which sedimentation rates were measured. Daily mean wind velocities and directions at Vlissingen (Royal
Dutch Meteorological Institute, KNMI) and Deurne (Royal Meteorological Institute of Belgium, KMI) were
used as proxy data for wave activity near the Paulina and Notelaar marsh respectively (Figure 1B).
These time-series of wind–wave and tidal activity, together with data on spatial factors like topographic
situation and vegetation cover, were analysed for influence on measured SSC and sedimentation rates, using
t-test procedures, one-way analysis of variance (ANOVA) and regression analysis. All statistical analyses were
performed using SAS/STAT software (SAS Institute Inc., 1989). Based on regression models, maps of the
spatial sedimentation pattern were computed in IDRISI (Eastman, 1994).
RESULTS
Exploratory data representation
In Figure 3 the distribution of sedimentation rates is summarized by boxplots, representing both the spatial
variations between the sampling sites and temporal variations between spring–neap cycles. Time-averaged
744 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 3. Whisker boxplots of sedimentation datasets, obtained by measurements over 25 spring-tidal cycles at 17 sampling sites. Theboxplots are plotted with respect to the position of sampling sites along the transects, situated as shown on Figure 2
SEDIMENTATION IN TIDAL MARSHES 745
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 4. Typical example of the temporal evolution of the inundation height above the marsh surface (in solid lines) and suspendedsediment concentration (SSC; in symbols and lines) during a tidal inundation at the Paulina marsh (thin lines and white symbols) and
Notelaar marsh (thick lines and black symbols)
Table II. P-values resulting from unpaired t-tests comparing initial suspended sediment concentration (ISSC) and sedimen-tation rate data (logSR) between winter and summer period at each measuring site (since variances of summer and winterdata sets are unequal, P-values using Satterthwaite’s computational method are presented here (SAS Institute Inc., 1989)).P-values in bold indicate that the difference between winter and summer period is not significant (α = 0·01) at these sites.For all other
sites there is a significant difference between winter and summer data
Data Site P Data Site P Data Site P
ISSC 1 <0·0001 logSR 6 0·0393 logSR 12 0·0235
10 <0·0001 7 0·0027 13 0·0525
logSR 1 0·0002 8 0·0003 14 0·00372 <0·0001 9 0·0009 15 0·00513 0·0082 10 0·0054 16 0·0400
4 0·0008 11 0·0020 17 0·0561
5 0·0041
sedimentation rates spatially range from 40 to 1650 g m−2 per spring–neap cycle, while at each sampling site
temporal variations are high, ranging in the order of 1 to 103 g m−2 per spring–neap cycle. The sedimentation
data sets are typified by skewed distributions, and are therefore first log transformed for each sampling site to
enhance normality for further t-tests and ANOVA.
Temporal patterns of sediment dynamics
During all sampled inundation cycles SSC is found to decrease with time, indicating that the suspended
sediment is continuously settling during the whole duration of inundation and that no resuspension occurs during
ebb tide (e.g. Figure 4). However the initial SSC (ISSC), measured at the beginning of marsh flooding, varies
considerably from one tide to another. Figure 5 shows that the ISSC linearly increases with maximum inundation
height, recorded at high tide, for all sampled inundation events, both at the Notelaar and Paulina marsh. In
addition this increase of ISSC with inundation height is greater during the winter period (October–March) than
during the summer period (April–September). An unpaired t-test showed that the difference in ISSC between
winter and summer is significant for both studied marshes (Table II).
The measured sedimentation rates vary between spring–neap cycles following a similar temporal pattern.
Figure 6 shows that sedimentation rates increase exponentially with cumulative inundation time. Especially for
inner marsh sites sedimentation rates are significantly higher during winter than during summer for the same
inundation time, while for sites situated next to creeks or marsh edges this seasonal difference is not always
significant (Figure 6, Table II).
Apart from tidal and seasonal influence, the role of wind–wave activity was examined. Table III shows that
for most sampling sites no significant relationship could be found between ISSC or sedimentation rates (SR) on
746 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
the one hand and average wind speeds on the other hand (see columns (b) and (e) in Table III). The relationship with
tidal inundation height or time is on the contrary highly significant for most sites (columns (a) and (d)). For these
sites the remaining variation, expressed by the residuals resulting from regression between ISSC/SR and inunda-
tion height/time, is also not significantly related to average wind velocity (columns (c) and (f)). Only for sites
15–17, situated on the Spartina marsh, sedimentation rate is not significantly related to inundation time and better
related to average wind velocity, especially for the winter period. This may be an indication that these marsh
sites, situated on the lower Spartina marsh bordering the marsh edge, are more sensitive to wind–wave activity.
Spatial sedimentation patterns
Figure 3 illustrates well that the spatial sedimentation pattern is influenced by the marsh surface topography.
A first topographic control is exerted by marsh surface elevation: low-lying marshes, such as the Spartina marsh
(sites 15, 16, 17), are characterized by much higher sedimentation rates than high marshes (sites 10 to 14), due
to more frequent, higher and longer inundations during the same spring–neap cycle (Figure 7a). However,
measuring sites which are situated next to tidal creeks or marsh edges do not follow this relationship. Only
when these sites are omitted from Figure 7a is a strong relationship found between sedimentation rate and
elevation.
A second topographic control is exerted by distance from the sediment source: along each sampling transect
sedimentation rates decrease with increasing distance from the creek or marsh edge (Figure 3). ANOVA con-
firms that sedimentation rates in inner marsh basins are not significantly different from each other, while they
Figure 5. Linear relationship between initial suspended sediment concentration (ISSC) and maximum inundation height observed at (a) theNotelaar marsh (site 1) and (b) the Paulina marsh (site 10). Note the difference between summer (April–September; indicated in whitesymbols and broken line) and winter (October–March; in black symbols and solid line) observations. Part (a) reprinted from Temmerman
et al. (2003), Figure 8, with permission from Elsevier Science
SEDIMENTATION IN TIDAL MARSHES 747
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 6. Exponential relationship between sedimentation rate (per spring–neap cycle) and cumulative inundation period (added up over allinundation cycles during a spring–spring cycle). Examples are shown for (a) a freshwater Phragmites australis marsh basin (site 1), (b) theadjoining natural levee (site 3), (c) a high salt marsh basin (site 10) and (d) the adjoining levee (site 12). Note the differences in axis.Summer (April–September) and winter (October–March) observations are plotted in white and black symbols respectively. The sampling
site locations are shown on Figures 2 and 3
are significantly higher on the levees (Table IV). Figure 7b shows that for all sampling transects the time-
averaged sedimentation rate, expressed relative to its value next to the creek or marsh edge, decreases exponentially
with increasing distance from the creek or marsh edge. However, sampling sites 16 and 17 do not fit this model
and are omitted, because sedimentation is here strongly influenced by the much lower marsh elevation. Our data
further suggest that absolute sedimentation rates are highest at the marsh edge and decrease along marsh creeks
with increasing distance from the marsh edge (Figure 7c).
The influence of the different marsh vegetation types on the spatial sedimentation pattern is illustrated by
the time-averaged sedimentation gradients, as measures for the efficiency of sediment trapping perpendicular to
marsh and creek edges. Surprisingly, these gradients are the same for all studied vegetation types (Figure 7b),
indicating that the intensity of sediment trapping is not influenced by the large differences in plant species,
height and growing density (Table I). ANOVA confirms that sedimentation rates in freshwater tidal marsh basins
with Phragmites australis or Salix vegetation are not significantly different (Table IV). The difference between
low (Spartina townsendii) and high salt marsh vegetation (mainly Puccinellia maritima, Aster tripolium and
Atriplex portulacoides) is significant (Table IV), but this is a consequence of difference in surface elevation
rather than in vegetation structure (Figures 3 and 7a). Also the significant difference between freshwater and
salt marsh basins (Table IV) may not be attributed to the vegetation cover, but to marsh topography.
748 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Table III. R2 and P-values resulting from linear regression analyses: (a) between initial suspended sediment concentration(ISSC) and maximum tidal inundation height (Ih); (b) between ISSC and daily mean wind velocity (Wd); (c) between theresiduals from regression (a) (ISSCRes) and Wd; (d) between sedimentation rate per spring–neap cycle (logSR) and cumulativetidal inundation time (It); (e) between SR and daily mean wind velocity averaged over the whole spring-neap cycle Wsn;(f ) between the residuals from regression (d) (SRRes) and Wsn. All regression analyses (a to f) are carried out for winter andsummer data separately. Notice that for most sampling sites significant relationships (P < 0·05) are found only between ISSC/logSR and tidal inundation height/time (indicated in bold). Only for sites 15–17, situated on the low Spartina marsh near
the marsh edge, is sedimentation rate better related to mean wind velocity
ISSC summer data ISSC winter data
(a) (b) (c) (a) (b) (c)Site ISSC × Ih ISSC × Wd ISSCRes × Wd ISSC × Ih ISSC × Wd ISSCRes × Wd
R2 P R2 P R2 P R2 P R2 P R2 P
1 0·42 <<<<<0·01 <0·01 0·75 0·03 0·21 0·56 <<<<<0·01 0·04 0·20 <0·01 0·7010 0·38 <<<<<0·01 0·07 0·34 0·14 0·15 0·16 0·04 0·02 0·43 <0·01 0·79
SR summer data SR winter data
(d) (e) (f) (d) (e) (f)Site logSR × It SR × Wsn SRRes × Wsn logSR × It SR × Wsn SRRes × Wsn
R2 P R2 P R2 P R2 P R2 P R2 P
1 0·35 0·07 0·02 0·78 0·03 0·73 0·54 <<<<<0·01 0·09 0·36 0·05 0·512 0·41 0·04 0·15 0·40 0·03 0·71 0·58 <<<<<0·01 0·09 0·36 0·01 0·743 0·71 <<<<<0·01 0·39 0·13 0·44 0·11 0·76 <<<<<0·01 0·18 0·19 <0·01 0·784 0·67 <<<<<0·01 0·13 0·42 <0·01 0·88 0·59 <<<<<0·01 0·28 0·10 0·06 0·485 0·59 <<<<<0·01 0·42 0·12 0·28 0·22 0·82 <<<<<0·01 0·20 0·16 0·10 0·366 0·77 <<<<<0·01 0·08 0·53 0·20 0·31 0·64 <<<<<0·01 0·19 0·18 0·02 0·677 0·85 <<<<<0·01 0·07 0·58 0·09 0·51 0·87 <<<<<0·01 0·16 0·23 0·02 0·658 0·70 <<<<<0·01 0·33 0·17 0·20 0·31 0·60 <<<<<0·01 0·05 0·51 0·13 0·299 0·84 <<<<<0·01 0·11 0·47 <0·01 0·86 0·76 <<<<<0·01 0·21 0·15 <0·01 0·8110 0·23 0·20 0·02 0·74 0·03 0·68 0·12 0·30 0·16 0·22 0·33 0·0711 0·46 0·04 <0·01 0·93 0·03 0·66 0·64 <<<<<0·01 0·02 0·64 0·17 0·1912 0·60 0·01 <0·01 0·98 <0·01 0·83 0·97 <<<<<0·01 <0·01 0·85 0·13 0·2513 0·28 0·15 0·21 0·22 0·18 0·26 0·86 <<<<<0·01 0·03 0·61 0·12 0·2714 0·59 0·01 <0·01 0·82 0·04 0·61 0·10 0·33 0·03 0·62 0·07 0·4115 0·43 0·05 0·41 0·07 0·55 0·02 0·08 0·37 0·47 0·01 0·34 0·05
16 0·16 0·29 <0·01 0·89 0·01 0·77 0·07 0·44 0·18 0·19 0·21 0·1517 0·07 0·52 <0·01 0·88 0·07 0·48 0·06 0·46 0·39 0·03 0·24 0·10
An integrated spatio-temporal model
The above-described analyses showed that spatial variations are partly explained by elevation differences,
which are in fact equivalent to differences in tidal inundation height and duration, which are the main factors
controlling temporal variations in sedimentation. It is then worthwhile to investigate to what extent both spatial
and temporal sedimentation patterns can be described in terms of a limited number of controlling parameters,
which interrelate and act synergistically.
A multiple non-linear regression model of the following form is proposed:
SR e e ec e = k lH mD nD (1)
where SR = the sedimentation rate (g m−2 per spring–neap cycle), H = the intensity of tidal inundation (this
parameter will be further specified below), Dc = the distance to the nearest creek or marsh edge (m) and De = the
SEDIMENTATION IN TIDAL MARSHES 749
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 7. (a) Mean sedimentation rate per spring–neap cycle in relation to marsh surface elevation (expressed relative to local mean highwater level) for all sampling sites. White dots indicate data points from levees (sites 3, 6, 7, 12) and are incorporated in the constructionof the broken regression line but omitted in the construction of the solid regression line. (b) Relative mean sedimentation rate in relationto distance from the creek or marsh edge for all sampling transects situated within different vegetation types (see Figure 2 for location ofthe transects). (c) Absolute mean sedimentation rate next to creek or marsh edges in relation to distance from the marsh edge, measured
along the creek system
distance to the marsh edge (m), measured along the nearest creek. For sampling transects perpendicular to the
marsh edge, De is set to zero. The model parameters k, l, m and n are determined using a non-linear regression
procedure in SAS/STAT (SAS Institute Inc., 1989).
First, regression analysis was carried out with sedimentation rates averaged over the one-year measuring
period as the dependent variable and incorporating the spatial variation between all salt and freshwater marsh
750 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Table IV. One-way ANOVA results for intercomparison between measuring sites during the winter, summer and wholeone-year measuring period. Pr > F-values in bold indicate that the difference in ISSC or sedimentation rate (SR) between
compared sites is not significant (α = 0·05). In all other cases there is a significant difference between sites
Data type Intercomparison between Compared sites Pr > F
Winter Summer Year
SR Fresh basins 1, 2, 5, 8, 9 0·3733 0·3074 0·4236
Fresh basins & levees 1, 2, 5, 8, 9, 3, 6, 7 0·0739 0·0160 0·0278
Salt basins 10, 11, 13, 14 0·1904 0·6411 0·6881
Salt basins & levees 10, 11, 13, 14, 12 0·0042 0·2853 0·0236Salt high & low marsh 10 to 17 <0·0001 <0·0001 <0·0001
Fresh & salt basins 1, 2, 5, 8, 9, <0·0001 0·0165 <0·000110, 11, 13, 14
ISSC Fresh & salt basin 1, 10 <0·0001 <0·0001 0·0028
Table V. Model parameters (k, l, m, n) and R2-values resulting from multiple non-linear regression analyses using Equation1 (see text)
SR data k l m n R2
Whole year average per site 1174·9 −0·3165 −0·0195 −0·0058 0·95Winter average per site 1565·5 −0·3352 −0·0174 −0·0046 0·93Summer average per site 737·3 −0·2036 −0·0298 −0·0137 0·98Winter all data 275·8 0·3006 −0·0216 −0·0043 0·68Summer all data 129·6 0·3943 −0·0392 −0·0075 0·56
sites. In this case H is estimated by surface elevation relative to local mean high water level. Figure 8a compares
sedimentation rates as observed and estimated by the regression model. It can be seen that the model is able to
predict almost all of the observed spatial variability (R2 = 0·95) without considering the large differences in
vegetation structure between the salt and freshwater marsh sites.
Secondly, a similar regression analysis was carried out, taking the distinction between winter and summer
sedimentation into account. Again, observations and model predictions are in good agreement (Figure 8a;
R2 = 0·93 and 0·98 for winter and summer respectively). The model parameter k is larger for the winter than for
the summer period, indicating that sediment input is larger during winter (Table V). The parameter l is more
negative during winter, which means that elevation differences have then a more pronounced effect on variations
in sedimentation rate. Seasonal differences in m and n values suggest that sedimentation gradients along and
perpendicular to tidal creek edges are less pronounced during winter than during summer. This confirms that
sediment trapping during flooding from the creeks to the inner marshes is greater during summer and conse-
quently less sediment reaches the inner marsh basins (see also Figure 6).
Finally, it was also investigated whether both temporal and spatial variations between spring–neap cycles
and between sampling sites can be modelled using Equation 1. In this case H is estimated by cumulative inunda-
tion height and is both time- and space-dependent. Figure 8b shows that the presented model structure can
partly explain the observed spatio-temporal sedimentation pattern (R2 = 0·68 and 0·56 for winter and summer
respectively).
Based on the regression models it is now possible to generate maps of the fully two-dimensional sedimen-
tation pattern in a raster-based geographical information system (GIS). For each raster cell that represents the
marsh surface a value of H, Dc and De has to be calculated. This was done for a raster image with 1 m by 1 m
cells of the Paulina marsh, where elevation data are available from airborne laser altimetry conducted by the
Dutch Rijkswaterstaat Meetkundige Dienst (minimum density 1 point/16 m2, guaranteed minimal vertical accu-
racy of 0·20 m) (Van Heerd and Van ’t Zand, 1999). From these elevation points a digital elevation model was
SEDIMENTATION IN TIDAL MARSHES 751
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 8. Comparison of sedimentation rates per spring–neap cycle as measured and predicted using multiple non-linear regression (seetext). (a) The model incorporates only the spatial sedimentation pattern by using for all sampling sites sedimentation rates averaged overthe whole one-year measuring period, over the winter and over the summer period. (b) The model incorporates both the spatial and temporalsedimentation pattern by using sedimentation rates as measured at each sampling site and for each individual spring–neap cycle during the
year. Regression was carried out separately for all winter and all summer data
computed using a triangulation with linear interpolation method. The tidal creek network was digitized based
on georeferenced recent aerial photographs and converted to a raster image. For each marsh surface cell, the
distance to the nearest tidal creek cell (Dc in Equation 1) and the distance of this nearest tidal creek cell to the
creek mouth at the marsh edge, measured along the creek (De in Equation 1), was calculated using the program
modules DISTANCE, COST and ALLOCATE in IDRISI (Eastman, 1994). Finally the sedimentation rate in
every marsh surface cell was calculated by solving Equation 1 and using the calculated values of H, Dc and De.
Maps of the whole year averaged and summer and winter averaged sedimentation rate per spring–neap cycle
were made using the appropriate model parameters in Table V. Figure 9a clearly shows that the calculated
sedimentation pattern is the result of the combined influence of surface elevation, the creek network, and
752 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Figure 9. (a) Simulation of the spatial sedimentation pattern at the Paulina marsh averaged over one year (in g/m2/spring–neap cycle).(b) Comparison between sedimentation rates per spring–neap cycle as measured and predicted after implementation of the regression model
in a GIS (see text) for sedimentation rates averaged over one year, over the winter and over the summer period
SEDIMENTATION IN TIDAL MARSHES 753
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
distance to the marsh edge. Measured and calculated sedimentation rates at the measuring sites are in very good
agreement (Figure 9b), indicating that the proposed method is very useful to compute two-dimensional patterns
of tidal marsh sedimentation.
DISCUSSION AND CONCLUSIONS
As reported from earlier studies, short-term temporal sedimentation patterns are in some tidal marshes mainly
controlled by wind–wave activity (Reed, 1989; Leonard et al., 1995; Van Proosdij et al., 2000), while other
studies indicate that tidal influence is more dominant (Allen and Duffy, 1998a; Christiansen et al., 2000). In the
meso- to macrotidal Scheldt estuary the tide is the most important factor that governs temporal patterns of tidal
marsh sedimentation.
For almost all sampling sites we observed an exponential increase of marsh sedimentation with increasing
inundation time. The exponential nature of this relationship can be explained as a consequence of a linear
increase in ISSC with maximum inundation height. Based on numerical modelling, Temmerman et al. (2003)
showed that the relationship between sedimentation rate and inundation time is exponential when ISSC increases
linearly with inundation height, while the relationship between sedimentation rate and inundation time is linear
when ISSC is assumed to be constant. The fact that an exponential relationship between sedimentation rate
and inundation time is found for most sampling sites suggests that the linear increase of ISSC with increasing
inundation height is a general mechanism that controls suspended sediment supply to the marsh surface.
Seasonal variations in sedimentation rates on tidal marshes are reported by several authors (e.g. Hutchinson
et al., 1995; Leonard et al., 1995; Leonard, 1997). Higher sedimentation rates are often found during the summer
period, which is explained by higher bioturbation of bottom sediments, leading to higher SSC and tidal marsh
sedimentation rates. However, we observed higher overmarsh SSC and sedimentation rates during the winter
period for the same inundation height or time. This is in accordance with the higher SSC values observed in
the stream channel of the Scheldt during the winter period (Fettweis et al., 1998).
The difference between winter and summer sedimentation rates is most important in inner marsh basins, while
this seasonal difference is not significant near creek and marsh edges. One possible explanation could be that
sediment trapping along flow paths from creeks to inner marshes is enhanced during the summer period, as a
consequence of higher growing densities and hydraulic resistance by marsh plants during summer (see also
Boorman et al., 1998) or by higher settling velocities of the suspended sediment during summer, for example
due to enhanced bioflocculation. Consequently less sediment is reaching the inner marsh basins during the
summer period. At this moment, however, quantitative data are lacking and further research is needed to fully
understand this seasonal sedimentation pattern.
Our study shows that the spatial depositional pattern on the tidal marshes along the Scheldt estuary can be
predicted from three morphometric variables only. As has been widely reported from other salt marshes, sedi-
mentation rates decrease with increasing surface elevation (e.g. Stoddart et al., 1989; Cahoon and Reed, 1995)
and with increasing distance from tidal creeks (e.g. French and Spencer, 1993; French et al., 1995; Leonard,
1997; Reed et al., 1999). Our study further shows that sedimentation rates along creek edges decrease with
increasing distance from the marsh edge, measured along the creek system. While former studies focused on
the identification of the different possible controlling variables, our study clearly shows that both spatial and
temporal sedimentation patterns can be well predicted using a single multiple regression model that only incor-
porates the three controlling variables discussed above.
The fact that the same model very well predicts the sedimentation patterns and rates on a salt and freshwater
marsh, located at the extremes of the estuarine gradient, suggests that the physical–sedimentological processes
controlling tidal marsh sedimentation are similar over the whole estuarine gradient of the Scheldt estuary. The
differences in vegetation characteristics, which strongly vary between and within the studied salt and freshwater
marshes, seem to have no detectable influence on the spatial sedimentation pattern. Marsh vegetation of course
reduces tidal currents and therefore promotes sediment deposition (Leonard et al., 1995; Leonard and Luther,
1995). However, it seems that very high and dense vegetation, which is typical for a freshwater Phragmites
australis or Salix marsh, is not more effective in tempering flow speeds and trapping sediments than typically
lower salt marsh plants such as Puccinellia maritima and Atriplex portulacoides.
754 S. TEMMERMAN ET AL.
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
In former studies, two-dimensional sedimentation patterns were calculated from a spatial network of meas-
uring sites, using conventional spatial interpolation techniques like kriging (French and Spencer, 1993; Leonard,
1997) and bilinear interpolation (French et al., 1995). French et al. (1995) especially emphasized the difficulties
of calculating sedimentation rates based on interpolation of spatially distributed measurements. In this regard,
this paper presents an alternative method to calculate two-dimensional spatial sedimentation patterns, by spa-
tially implementing an empirical model that takes into account the physical variables that determine the spatial
distribution of sediment over the marsh surface. Inundation frequency, height and duration are reflected in
the model by surface elevation, while the transport pathways of the sediment are reflected by the distance from
the creeks and the distance from the marsh edge measured within the creek system. Although the combined
influence of surface elevation and the creek network is successfully modelled, our approach also has certain
limitations. Especially where creek basins with an important difference in distance from the creek mouth are
adjacent, strong discontinuities in sedimentation rates may appear (Figure 9a). In order to handle such difficulties
a more hydrodynamically based model has to be used, which takes into account the complex flow of water and
suspended matter over the marsh surface topography and through the marsh vegetation cover.
ACKNOWLEDGEMENTS
This research was funded by the Institute for the Promotion of Innovation by Science and Technology in
Flanders (IWT), whose support is gratefully acknowledged. We also wish to thank all the people, and especially
Jos Meersmans, who assisted with the installation of the automatic measuring stations and during the fieldwork.
Digital elevation point data for the Paulina marsh were placed at our disposal by Rijkswaterstaat Meetkundige
Dienst and were used in this paper with permission.
REFERENCES
Allen JRL. 2000. Morphodynamics of Holocene salt marshes: a review sketch from the Atlantic and Southern North Sea coasts of Europe.Quaternary Science Reviews 19(12): 1155–1231.
Allen JRL, Duffy MJ. 1998a. Medium-term sedimentation on high intertidal mudflats and salt marshes in the Severn Estuary, SW Britain:the role of wind and tide. Marine Geology 150(1–4): 1–27.
Allen JRL, Duffy MJ. 1998b. Temporal and spatial depositional patterns in the Severn Estuary, southwestern Britain: intertidal studies atspring-neap and seasonal scales, 1991–1993. Marine Geology 146(1–4): 147–171.
Boorman LA, Garbutt A, Barratt D. 1998. The role of vegetation in determining patterns of the accretion of salt marsh sediment. InSedimentary Processes in the Intertidal Zone, Black KS, Paterson DM, Cramp A (eds). Geological Society, London, Special Publication139. Geological Society Publishing House: London; 389–399.
Cahoon DR, Reed DJ. 1995. Relationships among marsh surface topography, hydroperiod, and soil accretion in a deteriorating Louisianasalt marsh. Journal of Coastal Research 11(2): 357–369.
Cahoon DR, Marin PE, Black BK, Lynch JC. 2000. A method for measuring vertical accretion, elevation, and compaction of soft, shallow-water sediments. Journal of Sedimentary Research 70(5): 1250–1253.
Christiansen T, Wiberg PL, Milligan TG. 2000. Flow and sediment transport on a tidal salt marsh surface. Estuarine Coastal and Shelf
Science 50(3): 315–331.Claessens J, Meyvis L. 1994. Overzicht van de tijwaarnemingen in het Zeescheldebekken gedurende het decennium 1981–1990. Ministerie
van de Vlaamse Gemeenschap AWZ Afdeling Maritieme Schelde: Antwerpen.Cundy AB, Croudace IW. 1996. Sediment accretion and recent sea-level rise in the Solent, southern England: Inferences from radiometric
and geochemical studies. Estuarine Coastal and Shelf Science 43(4): 449–467.Eastman R. 1994. IDRISI for Windows 2·0 Users Guide. Clark University: Worcester, Mass.Fettweis M, Sas M, Monbaliu J. 1998. Seasonal, neap-spring and tidal variation of cohesive sediment concentration in the Scheldt Estuary,
Belgium. Estuarine Coastal and Shelf Science 47(1): 21–36.French JR, Spencer T. 1993. Dynamics of sedimentation in a tide-dominated backbarrier salt marsh, Norfolk, U.K. Marine Geology 110(3–
4): 315–331.French JR, Spencer T, Murray AL, Arnold NS. 1995. Geostatistical analysis of sediment deposition in two small tidal wetlands, Norfolk,
United Kingdom. Journal of Coastal Research 11(2): 308–321.Hoffmann M. 1993. Vegetatiekundig-ecologisch onderzoek van de buitendijkse gebieden langs de Zeeschelde met vegetatiekartering. Uni-
versity of Ghent: Ghent.Houtekamer NK. 1996. De schorren van de Westerschelde 1990/1993, overzichtskaarten van de vegetatie met begeleidende rapportage.
Rijkswaterstaat Meetkundige Dienst: Delft.Hutchinson SE, Sklar FH, Roberts C. 1995. Short term sediment dynamics in a Southeastern USA Spartina marsh. Journal of Coastal
Research 11(2): 370–380.Leonard LA. 1997. Controls of sediment transport and deposition in an incised mainland marsh basin, southeastern North Carolina. Wetlands
17(2): 263–274.
SEDIMENTATION IN TIDAL MARSHES 755
Copyright © 2003 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 28, 739–755 (2003)
Leonard LA, Luther ME. 1995. Flow hydrodynamics in tidal marsh canopies. Limnology and Oceanography 40(8): 1474–1484.Leonard LA, Hine AC, Luther ME. 1995. Surficial sediment transport and deposition processes in a Juncus-Roemerianus marsh, west-central
Florida. Journal of Coastal Research 11(2): 322–336.Meire P, Rossaert G, N. DR, Ysebaert T, Kuijken E. 1992. Het Schelde-estuarium: ecologische beschrijving en een visie op de toekomst.
Instituut voor Natuurbehoud: Hasselt.Neubauer SC, Anderson IC, Constantine JA, Kuehl SA. 2002. Sediment deposition and accretion in a mid-Atlantic (USA) tidal freshwater
marsh. Estuarine Coastal and Shelf Science 54(4): 713–727.Orson RA, Simpson RL, Good RE. 1990. Rates of sediment accumulation in a tidal freshwater marsh. Journal of Sedimentary Petrology
60: 859–869.Pasternack GB, Brush GS. 2001. Seasonal variations in sedimentation and organic content in five plant associations on a Chesapeake Bay
tidal freshwater delta. Estuarine Coastal and Shelf Science 53(1): 93–106.Pethick JS. 1992. Saltmarsh geomorphology. In Saltmarshes: morphodynamics, conservation and engineering significance, Allen JRL,
Pye K (eds). Cambridge University Press: Cambridge; 41–62.Reed DJ. 1989. Patterns of sediment deposition in subsiding coastal marshes, Terrebonne Bay, Louisiana: the role of winter storms.
Estuaries 12(4): 222–227.Reed DJ, Spencer T, Murray AL, French JR, Leonard L. 1999. Marsh surface sediment deposition and the role of tidal creeks: implications
for created and managed coastal marshes. Journal of Coastal Conservation 5: 81–90.Roman CT, Peck JA, Allen JR, King JW, Appleby PG. 1997. Accretion of a New England (U.S.A.) salt marsh in response to inlet migration,
storms, and sea-level rise. Estuarine Coastal and Shelf Science 45(6): 717–727.SAS Institute Inc. 1989. SAS/STAT User’s Guide, Version 6, Fourth Edition, Volume 1. SAS Institute Inc.: Cary, NC.Stoddart DR, Reed DJ, French JR. 1989. Understanding salt marsh accretion, Scolt Head Island, north Norfolk, England. Estuaries 12(4):
228–236.Taverniers E. 2000. Zeescheldebekken: de afvoer van de Schelde in 1999. Ministerie van de Vlaamse Gemeenschap AWZ Afdeling
Maritieme Schelde: Antwerpen.Temmerman S, Govers G, Meire P, Wartel S., 2003. Modelling long-term tidal marsh growth under changing tidal conditions and suspended
sediment concentrations, Scheldt estuary, Belgium. Marine Geology 193(1–2): 151–169.Van Damme S, De Winder B, Ysebaert T, Meire P. 2001. Het ‘bijzondere’ van de Schelde: de abiotiek van het Schelde-estuarium. De
Levende Natuur 102(2): 37–39.Van den Bergh E, Huiskes A, Criel B, Hoffmann M, Meire P. 2001. Biodiversiteit op de Scheldeschorren. De Levende Natuur 102(2): 62–
66.Van Eck GTM, De Pauw N, Van Langenbergh M, Verreet G. 1991. Emissies, gehalten, gedrag en effecten van (micro)verontreiningingen
in het stroomgebied van de Schelde en het Schelde-estuarium. Water 60: 84–99.Van Heerd RM, Van ’t Zand RJ. 1999. Productspecificatie Actueel Hoogtebestand Nederland. Rijkswaterstaat Meetkundige Dienst: Delft.Van Proosdij D, Ollerhead J, Davidson-Arnott RGD. 2000. Controls on suspended sediment deposition over single tidal cycles in a
macrotidal saltmarsh, Bay of Fundy, Canada. In Coastal and Estuarine Environments: sedimentology, geomorphology and geoarchaeology,Pye K, Allen JRL (eds). Geological Society: London; 43–57.
Wartel S. 1973. Variations in concentration of suspended matter in the Scheldt estuary. Bulletin of the Royal Belgian Institute for Natural
Sciences 49(2): 1–11.Wartel S. 1977. Composition, transport and origin of sediments in the Schelde estuary. Geologie en Mijnbouw 56(3): 219–233.